This invention relates to a magnetoresistance effect element, a magnetic head, a magnetic reproducing apparatus, and a magnetic memory and more particularly, to a magnetoresistance effect element which has a structure where a sense current is passed perpendicularly to a film plane of the magnetoresistance effect film, and to a magnetic head using the same, a magnetic reproducing apparatus and a magnetic memory.
By discovery of the Giant MagnetoResistance effect (GMR) in the laminated structure of magnetic layers, the performance of a magnetic device, especially a magnetic head is improving rapidly. Especially application to a magnetic head and MRAM (Magnetic Random Access Memory) of spin valve film structure (Spin-Valve: SV film) brought big technical progress to the magnetic device field.
A “spin valve film” has a structure which sandwiches a non-magnetic layer between two ferromagnetic layers. The magnetization of one ferromagnetic layer (called a “pinned layer” or a “magnetization pinned layer”, etc.) is fixed by an antiferromagnetic layer etc., and the magnetization direction of another ferromagnetic layer (called a “free layer”, a “magnetization free layer”, etc.) is rotatable in a response to an external magnetic field. And when the relative angle of the magnetization direction of a pinned layer and a free layer changes, a giant magnetoresistance change is obtained.
A CPP (Current-Perpendicular-to-Plane) type magnetoresistance effect element which passes sense current to a perpendicular direction to a film plane of such a spin valve film shows the still larger GMR effect compared with a conventional CIP (Current-In-Plane) type magnetoresistance effect element which passes sense current in parallel to a film plane.
On the other hand, a TMR element using the TMR effect (Tunneling MagnetoResistance effect) is also developed, which passes the current to a perpendicular direction as the CPP type GMR elements. However, in a TMR element, since the tunnel effect is used, when thickness of the non-magnetic intermediate layer made of alumina, for example, is made thin, there is a problem that MR rate of change decreases rapidly.
In the case of a TMR element, in the state where alumina is not made thin, resistance is very high. When application to a magnetic head is considered, its adoption is difficult from a viewpoint of a shot noise and a high frequency response. For example, in order to use for a magnetic head, AR (current passed area×element resistance) must be set to 1 Ωμm2 or less. However, in the case of a TMR element, there is a problem that MR rate of change disappears in this resistance level.
On the other hand, a CPP type magnetoresistance effect element has the advantage which has larger MR rate of change compared with a CIP type magnetoresistance effect element. In the case of a CPP type element, resistance of an element is dependent on element area. Therefore, when the miniaturization of the element is carried out, it also has an advantage that the amount of resistance change increases. This advantage serves as a big merit in the present when the miniaturization of a magnetic device progresses. Therefore, a CPP type magnetoresistance effect element and the magnetic head using it are considered to be the major candidates for realizing storage density from 200 gigabits per square inch (200 Gbpsi) to one Tbits per square inch class.
In the case of a TMR element, since an insulator is used for an intermediate layer, element resistance becomes high too much. For this reason, if the miniaturization of the element area is carried out, originating in high resistance and causing shot noise generating peculiar to a tunnel phenomenon and high frequency response degradation will pose problems. For this reason, a means of realistic solution is not found in application of a TMR element in high storage density of 200 or more Gbpsi.
In MRAM, tolerance level of element resistance is comparatively wide compared with a magnetic head. It is thought that a TMR element is applicable to MRAM of a first generation. However, also in MRAM, the miniaturization of the element area is carried out with improvement in storage density, and it is expected that a problem that the resistance becomes too high comes out. That is, also in any of a magnetic head and MRAM, high resistance peculiar to a TMR element poses a problem with improvement in storage density.
On the other hand, in the case of a CPP element using a metal non-magnetic intermediate layer, since the element resistance is very small unlike the TMR elements, the amount of resistance change is small while MR rate of change is large. As a result, it is difficult to acquire a high reproduction output signal. And in the case of spin valve film structure where realization possibility is the highest, only a free layer and a pinned layer are provided as the magnetic layers. That is, compared with a case of the artificial lattice multilayer structure, thickness and interfaces which contribute to MR rate of change are both insufficient. For this reason, MR rate of change becomes remarkably small compared with a practical MR rate of change.
In order to solve a pan of this problem, by laminating an oxide layer for the CPP clement which used the metal non-magnetic intermediate layer, increase of element resistance is aimed at and the trial to raise the amount of resistance change as for the same MR rate of change is made (K. Nagasaka et al. The 8th Joint MMM-Intermag Conference, DD-10).
In the case of this method, a metallic low resistance area is established in pinholes in part of oxide layer, and it aims to obtain a high resistance by constricting the current. However, it is difficult to provide pinholes uniformly. Resistance varies largely especially in a storage density of 100 Gbpsi or more for which element size of about 0.1 micrometers is needed. For this reason, fabrication of stable CPP elements is difficult.
By this technique, an increase in large MR rate of change cannot be realized, but resistance is just adjusted. That is, though MR rate of change does not change, if AR is raised, it is expected that the amount AdR of resistance change expressed with the product of MR rate of change expressed with percentage and AR win improve. Since area which contributes to MR rate of change becomes small effectually, MR rate of change seen from the whole may increase.
However, since element size becomes small so that it becomes high storage density, the resistance demanded from a viewpoint of a shot noise and the high frequency response characteristic must be small, for example, a case of storage density of 200 Gbpsi, tolerance level of AR (current passing area×resistance) is from about one mΩμm2 to a few hundreds mΩμm2. On the other hand, in the case of 500 Gbpsi class storage density, AR must be less than 500 mΩμm2. This is because element resistance becomes large, when the element size accompanying improvement in storage density contracts. Thus, it is required that AR should be made small with improvement in storage density. Therefore, it is clear that there is a limit in an approach to increase AdR (current passing area×resistance change) by increasing AR while keeping MR at a fried value. That is, the essential improvement in the MR rate of change itself is needed with improvement in storage density.
In order to improve a situation, research of a half metal prospers aiming at the essential improvement in MR rate of change. Generally, it is defined as a “half metal” being a magnetic material with which only either of the densities of states of a up-spin electron and a down-spin electron exists near Fermi level. When an ideal half metal is realized, two states of an infinite resistance state and a low resistance state are formed corresponding to the two magnetization states of the pinned layer and the free layer of an anti-parallel state and a parallel state. Therefore, MR rate of change of infinite size is ideally realizable.
Such an ideal state may be unable to be realized in fact. However, if a difference of density of states of a up spin electron and a down spin electron becomes larger than conventional material, an increase of MR rate of change does not remain in improvement in about 2 times, but a rise of 3 times, 4 times, and still more nearly extraordinary fast MR rate of change is expected.
That is, unlike the conventional solution mentioned above, improvement in large MR rate of change becomes essentially possible. However, there is a big problem which obstructs utilization which is explained below in a half metal investigated intensively now.
That is, the following material can be mentioned as a half metal material investigated until now. CrAs of semiconducting materials, such as NiMnSb of the CrO2 and the Whistler alloy with rutile structures, such as Fe3O4 with spinel structure, LaSrMnO with perovskite structure, and LaCaMnO, ZnO, GaNMn. Many of these materials have a complicated crystal structure. For this reason, in order to form a high quality crystal, substrate heating to a high temperature or special film formation technique is required. There is a problem that these processes are not easy to carry out in a creation process of an actual magnetoresistance effect element. This is the first problem.
A problem mentioned above may be solved by improvement of film formation technology. However, there are the following problems as a still more essential problem. That is, any case of half metal material known until now, a limit of curie temperature (Tc: in the case of Ferro magnetism) and Neel temperature (Tn: in the case of ferrimagnetism or antiferromagnetism) is at most 400K (about 100 degrees in centigrade) Since temperature which shows half metal nature (here, it is defined as Thm) becomes the lower temperature side, there is a problem that material which shows half metal nature in room temperature is not yet found. This is the second problem.
Thus, if half metal nature is realizable only at low temperature, the application to a consumer product is completely impossible. In order to use it as an actual magnetoresistance effect element, half metal appearance temperature Thm must be at least 150-200 degrees in centigrade or higher. In order to make Thm high, it is required to make Tc or Tn higher. However, with material investigated until now, Tc or Tn beyond room temperature hardly exists. Intensive research is made in order to raise Tc and Tn, but a decisive solution which raises Tc or Tn in every material cannot be found.
There are the following problems as the third big problem. That is, even if a half metal is realized in a single layer film, when it is provided in a multilayered structure like a spin valve, there is a problem that half metal nature in a laminated film interface is lost. This is because band structures differ in a bulk state in an interface of the lamination structure. Although a half metal is realized in a single layer film, there is a problem that a half metal is unrealizable in an interface or the surface. In a part of magnetic semiconductor material (CrAs), there is a report that a half metal of high Tc was realized. However, generally in an interface of a semiconducting material and metal material, diffusion is intense. For this reason, it is very difficult for half metal nature to be made not to be lost in a junction interface.
When using these magnetic semiconductor material, it is desirable to also constitute a non-magnetic spacer layer from a semiconducting material, and the combination with metal material is not realistic. If a specific material in an interface layer is not laminated in the case of the Heusler alloy material, such as NiMnSb, it is pointed out that half metal nature cannot be essentially realized (G. A. de Wijs et al., Phys. Rev. B 64, 020402-1). This originates in half metal nature being lost in a laminated structure interface, since symmetry in band structure of a crystal collapses near the interface.
With a CPP element using the Heusler alloy, even if measured at 4.2K or less cryogenic temperature which is the temperature below Tc, only MR rate of change lower than a spin valve film formed with the usual metal is observed. This is based on above explained problem.
With spin valve film structure, it must essentially be made a laminated structure. Since half metal nature will be lost near the interface, it is meaningless to pursue material which half metal nature by using a single layer of a single crystal.
As other means, there is a method of using a half metal as a material of a spacer layer. Here, a “spacer layer” is a non-magnetic layer which divides the pinned layer and the free layer in the case of a CPP element. An improved result using a perovskite system oxide is reported. For example, although Tc and Thm are still low temperature, when measured at temperature below Tc, a TMR element realized quite larger MR rate of change than a spin valve film of the usual magnetic material (J. Z. Sun et al., Appl. Phys. Lett. 69, and 3266 (1996)). However, it is difficult to create the pinned layer, the spacer layer, and the free layer using material with a special crystal structure like perovskite. And the above-mentioned second problem that Tc is low temperature is still not solved at all.
Thus, in extension of research of a half metal studied intensively now, realization of a high MR rate of change is difficult even in low temperature. Even if it is realized, a still bigger breakthrough for realizing large MR rate of change at room temperature will be needed.